jeudi 25 juin 2015

Rendez-vous à Toulouse en mars 2016 !

Les prochaines journées annuelles du GdR MASCOT-NUM auront lieu l'an prochain à Toulouse, du 23 au 25 mars 2016.

Soutenance d'HDR d'Emmanuel VAZQUEZ le 15 juillet à 14h30

Emmanuel VAZQUEZ

présentera ses travaux pour l’obtention de

l’Habilitation à Diriger des Recherches

portant sur le sujet :

« Sequential strategies based on kriging »


Le mercredi 15 juillet à 14 H 30
en Amphi Boucherot

CentraleSupelec – campus de Gif-sur-Yvette

Vous êtes cordialement invités au pot qui suivra dans la salle de réunion du département Signaux & Statistique (A3.05)

Membres du Jury :

 * Fabrice Gamboa, Prof. Institut de Mathématiques de Toulouse
 * Josselin Garnier, Prof. Univ. Paris-Diderot
 * Bertrand Iooss, Ingénieur-Chercheur EDF R&D
 * Luc Pronzato, DR CNRS
 * Michèle Sébag, DR CNRS

Résumé :

Emmanuel Vazquez entered the Ecole Normale Supérieure de Cachan in 1997 and was awarded the Agrégation de Physique Appliquée in 2000. In 2001, he obtained a Diplôme d'Etudes Approfondies in Mathematics for Vision and Learning, from the Ecole Normale Supérieure de Cachan. He received the Ph.D. degree in 2005 from the Orsay-Paris XI University, with a thesis on kernel-based non-linear systems black-box modeling. Since 2004, he is working at Supélec, now CentraleSupélec, as Associate Professor.

His research work focuses on sequential Bayesian search strategies and the design and analysis of computer experiments. This work is driven by a question central to many industrial problems: how to optimize the performance of a system using numerical simulations?  In particular, when simulations are time-consuming, it becomes essential to consider optimization algorithms that use the information provided by the simulations as efficiently as possible. The idea of the Bayesian approach for optimization is to use a random process as a model of the function to be optimized. Then, the optimization is performed by making evaluations of the function in sequence, each evaluation being chosen in order to minimize a criterion that quantifies the expected loss, under the random process model, incurred by taking the best evaluation result collected so far instead of the true unknown optimum. Both theoretical and practical aspects are considered in this work.